Potential of Machine Vision in Industrial Maintenance
There are ongoing research activities in the areas of industrial maintenance and machine vision in Kemi-Tornio University of Applied Sciences. Early experiences show that there is potential in applying machine vision to applications of the measuring and preventive industrial maintenance. For some reason machine vision seems to be widely unutilised despite of its potential in maintenance applications areas. Nowadays machine vision can be applied to large scale of on-line measurement tasks. It can be stated that everything you can see and even some part of the unseen can be measured by using machine vision
Introduction
The Optical measurement laboratory (OML) of Kemi-Tornio University of Applied Sciences has almost ten years of experience on machine vision technology. Nowadays OML has much of expertise about applying visible light machine vision techniques to industrial and other type of applications. In addition knowledge concerning x-ray and thermal imagining based machine vision has cumulated from the beginning of the year 2005.
Industrial maintenance and its role in intelligence service business is the main topic in 1.5 year R&D project (ie. ‘Compus Maintenance’), which started in September 2006. In this project, approximately 6.5 man-years of research resourcing and fifteen enterprises are available to investigate the state-of-the-art of industrial maintenance service business in Kemi-Tornio area. Main target is to develop a future vision and technology roadmap for more effective and preventive maintenance service network based on the core competencies of different enterprises. One of the goals is to find out machine vision application potential in the area of industrial maintenance and to study the possibilities to integrate vibration analysis and machine vision technologies and develop a pilot scale system which can be used in condition monitoring related to the preventive maintenance.
There is a huge economical potential related to the advance in the preventive maintenance. It has been estimated that 1 – 2% increment in the overall equipment effectiveness (OEE) in Finnish industry would result in savings of several million Euros. And in this field the option is clearly in developing more advanced and reliable monitoring technologies and analysis to assist the preventive maintenance operations.
Aspects and thoughts about potential of machine vision technology as a measurement technique in preventive industrial maintenance and some modern machine vision research facilities of OML are presented in this article.
Machine vision as a measurement technology in industrial maintenance
According to our preliminary research, machine vision hasn’t been much used in preventive and measuring maintenance despite of the significant potential. Quite common machine vision like application is off-line thermal imagining, which can be used to detect early failure by measuring unusual thermal radiation emitting from the target objects. For example the faulty slow speed rolling bearing conducts thermal radiation to support structures of the bearing, which situation can be detected by using thermal camera rather than for example by using vibration analysis. The result of the measurement is a thermal image, but most of these applications don’t include actual automated image processing. Thus, these measurements don’t fulfil the definition of machine vision. In many cases comparable to the previous one the off-line thermal imagining can be automated by using on-line thermal cameras and machine vision techniques nowadays.
On the other hand, the visible light camera technology is widely utilized in different kinds of manual monitoring applications. In quite many cases the purpose of the camera monitoring is to monitor unusual happenings, which can cause failures, from the processes. The image of the camera is transferred to a control room where an operator analyzes manually the images from the screens on more or less regular basis. The camera technology is present here, but again this is not machine vision due to absence of automated image processing.
Among the measuring and preventive maintenance, plenty of applications for machine vision technology will arise in the near future. By using a little bit of imagination, the measuring techniques of measuring maintenance can be compared to human senses. One of the common measuring technologies in industrial maintenance is vibration measurements that can loosely be compared with human feeling and hearing senses. On the other hand, ultrasound and acoustical emission as measurement technologies remotely resemble human hearing, and measurements with infrared camera can be compared again with the feeling sense. All the mentioned techniques already have plenty of applications where the maintenance-related measurements are implemented as automated on-line measuring systems to detect the early failures in the target processes of the maintenance. Logically, machine vision technology can be said to be similar to human sight, which is often used by the maintenance operators for the manual monitoring of the processes. The machine vision has the same huge potential as the previously mentioned sense comparable measurement technologies. Though, it is rarely used in applications of measuring maintenance. It must be also noticed that the infrared and the x-rays are located outside of the visible light in electromagnetic spectrum, but almost the same machine vision basic techniques, especially image processing can be utilised also in these wavelengths. Thus, machine vision can be used in applications where the human sight couldn’t even perform, and still the technology meets the technical definition of machine vision. Therefore we can state that everything you can see and even some part of the unseen can be measured by using machine vision.
Potential of machine vision in industrial maintenance
How many potential maintenance-related tasks there are in the industry that can be solved with the help of continuous visual measuring, and how many of them are reasonable to be automated? By extending this question also into infrared and x-ray vision areas, we can imagine several applications where it would be at least technically possible to apply machine vision technology for the needs of the measuring maintenance.
As a first example, the Optical Measurement laboratory has developed a prototype for condition monitoring of mine elevator steel ropes. On-line machine vision system measures parameters in the rope (e.g. diameter, center position, steel fibre detection), which are supporting the scheduled manual rope inspection. In another example, a pilot system for the condition monitoring of the bio fuel feeding process was implemented based on x-ray machine vision. The goal was to detect impurities (rocks, metal pieces etc.) inside the bio fuel and monitor the condition of the bull chain conveyor. The results showed that xray based machine vision survived and succeeded in those tasks. As a third example, an infrared based machine vision prototype system for condition monitoring of polymer rollers was developed by Satakunta University of Applied Sciences and Metso Automation in 2002 /1/. A thermal camera was measuring surface temperature deviations in the roller. Detection of the hot spots can be used to predict the failures of the polymer rollers.
In addition, a very preliminary studies (Figure 1) and discussions from the field of oil analysis have been done in Optical measurement laboratory together with the industry and universities. It might be possible to perform on-line monitoring at the condition of the oil and especially the machines which are lubricated with the oil by using machine vision. There is huge potential in this area, but there are also many technical challenges ahead. Furthermore, quite tangible scenario for the near future seems to be on-line infrared machine vision systems in measuring and preventive maintenance. It is possible to predict failures or even wear and tear in several different devices by using automatic surface temperature measuring (thermal camera) and automated image processing. The technology of the infrared cameras is rapidly developing (resolution and frame rate increase, digital interfaces to PC become more common) and the prices of infrared cameras have been tremendously decreased in past few years. These things create many possibilities to infrared based machine vision application developers.
Technical reliability of machine vision, just like other “hitech” measurement devices used in automation, has been proven to be high in several demanding environments. Applications as log-shape measuring in sawmill-industry, assembly machines in electronics industry, web-edge tracking in processing industry, joint tracking in welding automation, and reverse vending machines that credit for empty bottles, can be presented as some references. The harsh environmental conditions in maintenance applications have to be carefully taken care of when applying machine vision to industrial condition measurements.
It is clear that in the aspect of fundamental purpose of measuring maintenance it is insignificant which technology is used for monitor the processes. Machine vision won’t solve all of the measuring maintenance tasks in the future, but there are and will be several application targets where the machine vision produces added value for the maintenance purposes. Potential exists also in linking machine vision with other measurement technologies. For example by linking measurements from vibration analysis and machine vision, it can be possible to achieve such measurement information that hasn’t been utilized before. This is one of the topics that will be further studied in the Compus Maintenance project.
Modern research facilities
In PAL project Optical measurement laboratory has designed and implemented a modern and quite unique research environment for the ongoing and for the future machine vision research projects. The core of the environment is aluminium profile mechanism (Figure 2). For measuring of static and moving target objects there are a table and a belt conveyor located under the mechanism. The machine vision components can be attached and adjusted in six degrees of freedom for achieving the optimal setup for the applications or research cases. A baggage scanner like x-ray system and a thermal camera are also part of this environment. Thus, the machine vision technology including the wave lengths outside the visible spectrum can be developed by using this environment.
The modern research environment can be used to research and test different kind of imagining geometries and systems by using wide scale of machine vision components under controlled circumstances. The importance of the appropriate imagining methods and components are noticed during the machine vision research years. It is very important that the machine vision system equipment design and testing are done carefully before the software (image processing) development. For example appropriate object illumination in visible light machine vision is a well known key issue for achieving high quality images or in thermal imagining applications the angle of the camera respect to surface of the target can be crucial. In many cases the different kind of concrete equipment based parameters of the imagining setups has to be tested before the actual application development or implementation.
The investment for the machine vision research facilities gives a strong basis for the development challenges of the machine vision based industrial maintenance applications among other application areas for the OML and the enterprise partners in the future.
References
/1/ Soini. A. Suomela. J. Aimonen. P. Public Final Report of EUSTIST-IMV Activity THERM –Continuous Infrared Measurements in Paper Industry, Loppuraportti, 2002
|
